Mission
Produce independent evidence about when AI strengthens scientific reasoning, when it weakens it, and how research teams can tell the difference.
ABOUT SANTAROS LABS
A Nonprofit Lab for Rigorous Human–AI Science
Santaros Labs is an independent nonprofit computational research lab. We investigate how AI changes the way scientific teams generate hypotheses, analyze data, write and review code, synthesize literature, and evaluate uncertainty.
Our Purpose and Direction
Vision
Scientific AI that expands human capability while preserving accountability, reproducibility, methodological pluralism, and expert judgment.
Nonprofit by Structure, Research-First by Design
Santaros Labs operates for scientific and public benefit rather than private distribution. Funding is directed to research staff, compute, data stewardship, independent review, replication, and dissemination.
We do not claim tax-deductible status, institutional accreditation, ethics approvals, or completed findings unless those details are verified and published for the relevant entity or study.
RESEARCH INTEGRITY
Responsible From Question to Claim
Protocol before outcome
We define the research question, comparison conditions, measures, exclusions, and stopping rules before interpreting results.
Risk-proportionate oversight
Human-participant work does not begin until the appropriate independent ethics review, consent process, and data controls are in place.
Traceable evidence
We preserve sources, transformations, code, model settings, and decision logs so claims can be audited and reproduced.
Built for Scientific Trust
Independent nonprofit mission
Santaros Labs is research-first and nonprofit. Resources support scientific work, infrastructure, and public-interest outputs.
Honest study status
Proposed, preregistered, recruiting, analyzing, and completed work are labeled separately. Plans are not presented as findings.
Reproducible by design
Study artifacts are prepared for reuse through versioned protocols, analysis environments, and machine-readable provenance.
Public-interest outputs
Where privacy, consent, licensing, and security permit, we publish methods, code, instruments, and negative results.
RESEARCH PARTNERSHIPS
Bring a Question Worth Testing
We welcome conversations with principal investigators, nonprofit institutes, universities, open-source communities, and funders working on trustworthy AI for science.
Methods, governance, and authorship expectations are discussed before a project begins.